Authorship at the Threshold: Generative AI between Instrumentality and Agency

Abstract

The rapid spread of generative artificial intelligence has reopened the philosophical and aesthetic debate on the concept of authorship. This article offers a theoretical map of the main positions in the existing literature, organized around three conceptual thresholds that structure our perception of authorship in AI-generated works. The first is the threshold of authorial relevance: not all cultural artifacts invite reflection on authorial intent; this threshold is crossed when a work seems to demand interpretive depth and points toward a creative consciousness behind it. The second is the threshold of instrumentality: as the causal distance between human input and generated output increases, it becomes increasingly difficult to attribute creative authorship to the user who formulated the prompt, raising the question of whether AI can be considered a co-author or even an autonomous author. The third is the metaphorical-literal threshold: this distinguishes the performative, knowingly metaphorical attribution of intentionality to AI (an anthropomorphizing “as-if” stance that facilitates interaction) from a literal belief that the system possesses genuine consciousness or agency. The article then examines the main theoretical positions on the issue, from exclusively human authorship to authentic artificial authorship and distributed conceptions of creativity, highlighting both their argumentative strengths and limitations. In conclusion, two key factors are identified as determining where one stands in relation to these thresholds: the degree of perceived creativity of the system (understood as the distance and unpredictability between input and output), and the level of affective and aesthetic engagement on the part of the audience, since we tend to “feel” the presence of an author above all when we are emotionally moved by a work.